Differential Privacy Noise Magnitude and LLaMA-2 Performance on Big-Bench Hard Tasks

This report synthesises findings from 7 peer-reviewed papers addressing the following research question: What is the correlation between the magnitude of differential privacy noise and the performance degradation of LLaMA-2 on specific Big-Bench Hard tasks requiring multi-step logical inference. 9 claims were extracted from source literature; 9 were independently verified against retrieved documents. An automated multi-reviewer quality assessment produced a score of 8.6/10. This report is a machine-generated literature synthesis and does not constitute original research.

Research goal: What is the correlation between the magnitude of differential privacy noise and the performance degradation of LLaMA-2 on specific Big-Bench Hard tasks requiring multi-step logical inference?

Autonomous literature synthesis. Automated review score: 8.6/10. Full text and citation available at Assignee Research.